Nov 15, 2026 09:20 8 reads

How to Read Valve Pressure Test Reports from Chinese Factories Without Getting Fooled

A factory in Wenzhou once sent me a pressure test report so clean it made me suspicious. Every valve passed, both directions, zero leakage, all signed. Then I noticed the report covered the lot but the test bench log showed 34 valves tested for a purchase order of 1,200. That is the thing about test reports out of the valve trade: a report is a document about a test, and the two are not always the same event.

Pressure testing is the one piece of quality evidence that a valve buyer can read without a metallurgy lab. So it is worth knowing exactly what the standard tests mean, what directions they run in, and what a report needs to contain before you can rely on it.

The three tests, and what each one is protecting you from

Nearly every industrial valve standard you will meet in China runs variants of the same three tests. ISO 5208 gives you leakage rate classes (A through G, with the strict metal seat classes running D to G in most industrial work). API 598 is the petroleum-side equivalent with its own wording. EN 12266-1 is the European one, also with rate classes.

The shell test, sometimes called the body test, pushes pressure through the valve with the closure open. It is normally 1.5 times the nominal pressure rating at 20°C for the duration the standard specifies (often 15 seconds to a few minutes depending on size). This test is looking for a burst or a weeping casting defect. It is the test that matters most for casting quality, and it is the one most likely to be skipped, because it requires blanking plates and time.

The seat test runs with the valve closed and pressure applied to one side. This is the leak you actually care about in service. Note carefully: nearly every standard requires it in both directions for bidirectional valves, and only in one designated direction for valves with a preferred sealing direction. That distinction is where most reports go soft.

The backseat test applies only to valves with a backseat, mostly globe and gate valves, and tests the seal between the stem and the bonnet that lets you repack the stem under pressure. It is frequently omitted from reports because a lot of samples do not have a functioning backseat even though the drawing shows one.

Four lines that tell you whether the test happened

When a report arrives, I go straight to these before I read anything else:

  • Test pressure in bar or psi, written as an actual number, with the calculation shown (for example 1.5x PN16 = 24 bar). If the report just says "per standard," the number was transcribed from a template.
  • Holding duration. Standards specify seconds to minutes; if the report has no duration column, nobody was watching a gauge.
  • Serial numbers or heat numbers per valve, mapped one to one against the delivery. A report that says "all valves pass" without a list is a marketing document.
  • Test medium. Water for shell tests, air for some seat tests, and the medium matters because a valve that passes on water can fail on air (air molecules are far smaller and find paths water does not).

If all four are present and consistent, the report is probably real. If two or more are missing, ask for the bench log with timestamps and the inspector's name, and see what happens to the response time. Slow answers are an answer.

The pressure direction trap

A butterfly valve with a soft seat, a check valve, and a ball valve all need their sealing direction specified, and the Chinese factory defaults vary. I have seen a batch of wafer butterfly valves tested only in the direction that flatters the seal, delivered, and then fail on the customer's line because the installation ran the other way and the pressure pushed the disc off its seat.

The fix is contractual, and it is one sentence: state that seat testing is to be carried out in both directions at the rated pressure and shall be recorded per unit. If a supplier tells you their product only seals one way, believe them and design around it, or buy a double-offset or trunnion design that does seal both ways. Do not let the report style decide your product.

Castings, and why the shell test is not enough

For cast bodies in CF8M or WCB, a shell test catches through-wall defects and nothing else. Porosity that sits below the surface will pass the shell test, survive painting and shipping, and then weep after a season of thermal cycling in service. This is why serious buyers pair the pressure test with either a radiograph or a dye penetrant check on the pressure boundary.

There is a negotiation trick that works well here. Ask the foundry or the valve factory to hold production pending review of the radiograph prints before machining. It costs the factory a few days of cash conversion, so they will push back, but the good ones accept it once and then build it into their scheduling. The ones who refuse outright are telling you their castings are not inspectable.

What the standards actually require on documents

If your end customer is European and asks for EN 10204 3.1 material certificates, understand what that is: a certificate issued by the manufacturer's own inspection department, not an independent body. 3.2 means an independent inspector signs off. Both are legitimate; they are not the same thing and they are not interchangeable in a specification.

The pressure test report is a separate document from the material certificate. I have had factories offer one as a substitute for the other, usually because the material certificate could not be obtained for the actual heat used. When that happens, the heat number on the casting is the thing to check against the delivery paperwork, and it is worth commissioning one XRF spot check per heat if you cannot get the paperwork.

A cheap way to verify at receipt

You do not need a hydrostatic bench to spot check. A hand pump with a gauge, two blind flanges and a bucket of water gets you a shell test at 1.5 times rating on a sample of five valves in about half an hour. Record gauge readings with your phone's video, and keep the video with the batch records. When a dispute arises later, that footage has resolved more arguments for me than any certificate.

Seat testing on a sample, in both directions, with a known reference pressure, plus a shell test at 1.5x, plus a per-unit list of serials, is the entire minimum. Everything else is refinement.

← More articles  |  Home